Does respect for autonomy imply respect for precedent autonomy? The principle of respect for autonomy requires us to respect a competent patient’s treatment preference, but not everyone agrees that it requires us to respect preferences formed earlier by a now‐incapacitated patient, such as those expressed in an advance directive. The concept of precedent autonomy, which concerns just such preferences, is problematic because it is not clear that we can still attribute to a now‐incapacitated patient a preference which that patient never (...) disaffirmed but can no longer understand. If we cannot make that attribution, then perhaps we should not respect precedent autonomy – after all, how can you respect patient autonomy by giving patients what they no longer want, even if they never disaffirmed those wants? I argue that whether an earlier preference can still be attributed to a now‐incapacitated patient depends on the reasons behind the preference, for a preference includes (and is not merely supported by) the reasons behind it. When the considerations that served as reasons no longer exist, neither does the preference which included those reasons. In particular, if the considerations that served as reasons for the patient exist only under conditions where the patient retains full mental capacity, then once that capacity is lost, so are those reasons and the preference based upon them. I use this analysis of precedent autonomy to ascertain the merits of various approaches to advance medical decisionmaking, including Nancy Rhoden’s approach, approaches based on a Parfitian personal identity analysis, approaches based on soft paternalism, and approaches based on the stability and longevity of preferences. Despite the apparent absurdity of respecting patient autonomy by giving patients what they no longer prefer but have never disaffirmed, I conclude with some programmatic remarks on when and why respect for (precedent) autonomy nonetheless requires us to respect former preferences. (shrink)